The oldest continuously operating bank in the U.S. is getting an upgrade. On Thursday at an internal town hall, BNY CEO Robin Vince unveiled the next-generation version of Eliza, its AI platform that allows employees to create bespoke agents and carry out tasks such as accelerated check processing built on reams of company data.
“We think of AI as generating capacity for us,” Vince said at the town hall. According to a BNY spokesperson, 98% of the company’s employees are trained on generative AI, with the majority using Eliza daily.
The announcement comes as the broader banking industry embraces AI, often through internally developed products such as BNY’s Eliza. Research firm Evident tracks the 50 top banks and, according to its October 2024 report, every firm on the index has referenced AI in at least one investor relations document. Evident ranked BNY as 14th in terms of AI adoption, with JPMorgan Chase and Capital One in the lead, though BNY’s standing is likely to change given its recent push.
The AI financial future
Named after the wife of Alexander Hamilton, who founded BNY’s corporate predecessor, Eliza has stood as the flagship AI product of the 240-year-old bank. The new update for Eliza, which first launched in 2024, will focus on allowing employees to access data across different sources, from different hubs within BNY to live market data, with the goal of increasing speeds for responding to clients.
In an email sent to employees and shared with Fortune, chief people officer Shannon Hobbs and chief information officer and global head of engineering Leigh-Ann Russell wrote that the upgrade would allow them to develop client briefings in minutes rather than hours.
The broader question for banks, which are constrained by compliance requirements and red tape, is whether AI tools will evolve beyond pilot programs and transform core functions. According to the Evident report, just six of its tracked banks reported that AI had had a financial impact.
BNY chief data and AI officer Sarthak Pattanaik previously toldFortune that the bank uses a “walled garden” approach to train its systems, though it uses external models from companies like Anthropic, Google, and OpenAI, and conducts red-team testing to simulate cyberattacks.
In their email, Hobbs and Russell referenced different use cases already adopted by bank employees for Eliza, including reconciling weekly transactions (which has increased automatic processing); repairing payments; and writing code. “Fluency in AI is no longer optional, it’s critical,” they wrote.